Working hour statistical method and system for engineering mechanical equipment and electronic equipment

Abstract

The invention relates to the technical field of man-hour statistics, in particular to a man-hour statistics method and system for engineering mechanical equipment and electronic equipment, and the method comprises the steps: taking an ACC switching signal as a triggering condition, and starting statistics only when the signal is turned on; the method comprises the following steps: synchronously collecting multi-sensor data, extracting sliding window statistical characteristics, outputting a state result and confidence through a trained decision model, checking and eliminating interference through an interference characteristic library, processing state conflicts and undetermined states in combination with a misjudgment elimination rule, accumulating continuous working time exceeding an exclusive working time threshold value as effective working time, and finally, determining the working time. And the model parameters are regularly adjusted and optimized and the interference feature library is updated. According to the invention, through multi-sensor data fusion and intelligent determination, state results are accurately distinguished, and the man-hour statistical accuracy is improved; through verification of an interference feature library and misjudgment elimination, environmental interference and artificial counterfeiting are avoided; and through adaptive updating and adaptation to various types of equipment and complex working conditions, reliable data is provided for construction accounting.

Description

Technical Field

[0001] This invention relates to the field of working time statistics technology, and in particular to a method, system and electronic equipment for calculating working time of engineering machinery and equipment. Background Technology

[0002] Construction machinery and equipment are widely used in earthwork, road construction, and other projects, and their actual operating time directly affects construction costs and efficiency. Traditional timekeeping usually relies on manual recording or single sensor monitoring; manual recording is prone to problems such as omissions and false reporting; early mechanical timers could only monitor single parameters such as engine speed, and their monitoring accuracy was low, their adaptability was poor, and they could not effectively solve the problem of timekeeping.

[0003] Currently, some devices such as Beidou and GPS locators use ACC signals to determine engine start-up status to calculate working time. However, ACC only reflects engine status and cannot distinguish between idling and actual operation, nor can it resist ambient vibration interference or human intervention. On the other hand, relying solely on single vibration, speed, or tilt angle data under complex operating conditions can easily lead to misjudgments or omissions. Existing research shows that multi-sensor fusion can significantly improve the reliability and accuracy of detection results, but existing construction machinery time monitoring systems mostly use simple threshold judgments, lacking intelligent analysis and adaptive capabilities, making it difficult to meet the needs of eliminating interference and preventing data falsification.

[0004] Therefore, there is an urgent need for a time statistics method that can improve the accuracy of time monitoring, eliminate interference, and have the ability to adaptively adjust to different types of construction machinery and equipment. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, and electronic device for calculating working hours of engineering machinery and equipment, so as to solve the problems of inaccurate equipment status identification, weak anti-interference ability, and poor equipment adaptability in the prior art.

[0006] The technical solution of this invention is: a method for calculating the working hours of engineering machinery and equipment, comprising: Collect the ACC switch signal of the construction machinery equipment. If it is in the on state, perform the following steps: Data from multiple sensors is collected and processed synchronously according to the sampling frequency, and statistical features within the sliding window are extracted from the processed multi-sensor data according to a preset sliding window. The statistical features are input into the trained decision model, and the state results and confidence scores of each sliding window are output; the state results include the device state and the undetermined state, and the device state includes the working state and the non-working state. The state results of each sliding window are verified based on the interference feature library. If interference is found, the sliding window is determined to be in a non-working state, and its confidence level is adjusted to a preset high confidence threshold. If there is no interference, the state result of the sliding window is not changed. The status result of each time point is marked based on the status result of the sliding window. According to the misjudgment elimination rule, the conflict of status result and the pending status of each time point are handled. The time points that are continuously marked as working status are accumulated. When the accumulated time is greater than the exclusive working time threshold of the engineering machinery equipment, it is included in the effective working time. Regularly assess the accuracy of the status results, and adjust the parameters of the decision model and update the interference feature library based on the feedback.

[0007] Preferably, the method for verifying the state results of each sliding window based on the interference feature library includes: The system calls a preset interference feature library and compares the statistical features extracted from the sliding window with the interference features in the interference feature library. If the statistical features of the sliding window match the interference features, it is determined that the sliding window has interference; otherwise, it is determined that there is no interference.

[0008] Preferably, the interference feature library pre-stores interference features caused by human fraud and external environmental interference, and the dimensions of the interference features are consistent with the statistical features extracted by the sliding window.

[0009] Preferably, the method for comparing the statistical features extracted by the sliding window with the interference features in the interference feature library is as follows: The statistical features are normalized and then similarity is calculated with the interfering features; the similarity calculation uses cosine similarity or Euclidean distance algorithms. If the calculated similarity value is greater than or equal to the preset similarity threshold, or the Euclidean distance value is less than or equal to the preset distance threshold, then the statistical features of the sliding window are determined to match the interference features.

[0010] Preferably, the method for handling conflicting state results and pending states at each time point according to the misjudgment elimination rule includes: For each time point, summarize all state results covering that time point and their corresponding confidence levels, and use the confidence level voting method to determine the state result for that time point; If there are consecutive time points that are all marked as pending, and the length of the corresponding time period is greater than or equal to the preset pending time length threshold, then the device status of all time points within that time period will be uniformly corrected to non-working status. If there are single or consecutive time periods shorter than the pending time duration threshold that are marked as pending, the device status is determined based on the preceding and following time periods: When the device status is consistent between the current and subsequent time periods, the device status at multiple time points within a single or continuous time period is set to be consistent with the device status of the current and subsequent time periods. When the device status is different between the current and subsequent time periods, all status results covering the current time point or each time point within a continuous time period and their corresponding confidence levels are summarized, and the confidence level voting method is used to determine the device status of the pending state time point or the pending state time point within a continuous time period.

[0011] Preferably, the confidence voting method includes: for each time point, accumulating the confidence scores corresponding to each state result to obtain the sum of the confidence scores corresponding to each state result, and marking the time point with the state result that has the highest sum of confidence scores.

[0012] Preferably, the specific working time threshold for the engineering machinery equipment is determined based on the equipment type, operating characteristics, and preset working condition requirements, and is used to provide a benchmark for determining effective working time.

[0013] On the other hand, this application also discloses a working time statistics system for engineering machinery and equipment, including: The ACC signal prior module is used to determine the operating status of the equipment based on the ACC switch signal, thereby determining the start and stop logic for subsequent working time statistics. The data acquisition and processing module is used to synchronously acquire and process multi-sensor data according to the sampling frequency, and extract statistical features within the sliding window from the processed multi-sensor data according to a preset sliding window. The decision module is used to input statistical features into the trained decision model and output the state results and confidence scores of each sliding window; the state results include the device state and the pending state, and the device state includes the working state and the non-working state. The interference verification module is used to verify the state results of each sliding window based on the interference feature library. If interference is found, the sliding window is determined to be in a non-working state. If there is no interference, the state result of the sliding window is not changed. The working time statistics module is used to mark the status result of each time point based on the status result of the sliding window, handle the conflict of status results and pending status at each time point according to the misjudgment elimination rules, accumulate the time points continuously marked as working status, and when the accumulated time exceeds the exclusive working time threshold of the construction machinery equipment, it is counted as valid working time. The adaptive update module periodically evaluates the accuracy of the state results, optimizes the parameters of the decision model based on feedback, and updates the interference feature library.

[0014] On the other hand, this application further discloses an electronic device, including: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a time statistics method for engineering machinery equipment as described above.

[0015] Compared with the prior art, the advantages of the present invention are: (1) This invention uses the ACC signal as a priori trigger for the working time statistics process, filtering out invalid interference from non-engine operating states at the source. Then, it collects data through a multi-sensor set and extracts multi-dimensional statistical features using a sliding window to comprehensively capture the parameter differences between working and non-working states. The trained decision model outputs state results with confidence, and combined with the specific working time threshold for engineering machinery equipment, only the periods when the continuous working time exceeds the threshold are counted as valid working time. This design breaks through the limitations of traditional single signals or simple thresholds, accurately separating invalid time such as idling and idle speed, so that the statistical results are highly consistent with the actual workload of the equipment, providing accurate data support for cost accounting and efficiency evaluation.

[0016] (2) This invention constructs a dual protection system through an interference feature library and a false judgment elimination rule. The interference feature library pre-stores multi-dimensional feature vectors of human-induced falsification and environmental interference. By matching feature vectors, the equipment status of the matching window is determined. For conflicts and pending states at time points, a confidence-based voting method is used to accumulate scores to determine the state at the time point, and pending states that exceed a preset duration are corrected to non-working states. This mechanism comprehensively shields false signals such as instantaneous interference and human-induced shaking, significantly reducing the proportion of false working hours and providing an objective basis for construction settlement and equipment assessment.

[0017] (3) The present invention uses an adaptive update module to periodically evaluate the judgment results, dynamically optimize the decision model parameters and update the interference feature library to adapt to changes in equipment operating status and new interference features; at the same time, the dedicated working time threshold can be flexibly adjusted according to the equipment type and operation characteristics, without the need to customize the system for each type of equipment. When the system of the present invention is deployed to new equipment, it can quickly complete the adaptation through historical data, which not only reduces deployment and maintenance costs, but also ensures the consistency of statistical accuracy under different models and operating conditions, and significantly improves the universality and scalability of the system in various types of equipment in engineering sites. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of a method for calculating the working hours of engineering machinery equipment according to the present invention; Figure 2 This is a structural block diagram of a working time statistics system for engineering machinery equipment according to the present invention. Detailed Implementation

[0019] The present invention will be further described in detail below with reference to specific embodiments: This application is applicable to various engineering scenarios such as earthwork excavation, road construction, building construction, and mining. It is compatible with different types and models of construction machinery and equipment such as excavators, loaders, and road rollers. These scenarios generally have problems such as complex working environments, variable working conditions, and high risks of human error in reporting working hours, which traditional working hour statistics methods cannot handle. This application adopts a technical solution that is deeply integrated with the needs of the scenario through ACC signal prior triggering, multi-sensor synchronous acquisition and sliding window feature extraction, decision model judgment, interference feature library verification, misjudgment elimination rules and adaptive updates. It solves three major technical problems in existing technologies: the inability to accurately distinguish between effective work and invalid states, resulting in distorted working hour statistics; insufficient resistance to environmental interference and prevention of human error; and poor adaptability to different types of equipment. It fully meets the actual needs of accurate working hour calculation on engineering sites, ensures the authenticity and reliability of data, and adapts to multiple equipment and multiple scenarios.

[0020] This application provides a method for calculating the working hours of construction machinery and equipment, including: Collect the ACC switch signal of the construction machinery equipment. If it is in the ON state, perform the following steps, such as... Figure 1 As shown, it includes: S1. Collect and process multi-sensor data synchronously according to the sampling frequency, and extract statistical features within the sliding window from the processed multi-sensor data according to the preset sliding window.

[0021] Specifically, the ACC switch signal is a switch signal used in construction machinery to control the on / off state of the engine's auxiliary current. Its core function is to reflect the engine's start / stop status. When the operator turns the key or presses the start button to the ACC position, the signal is activated, indicating that the engine has started. When the key or button is turned off, the ACC switch signal is deactivated, indicating that the engine has stopped running. The engine's ACC switch status is used as a priori condition for determining the operating status; sensor data is only further analyzed after ACC is activated. When ACC is deactivated, the equipment is quickly set to a non-operating state to avoid misstating vibrations occurring when ACC is off.

[0022] Multiple sensors are deployed at various key parts of the engineering machinery equipment to comprehensively collect multi-dimensional operating condition data during equipment operation. The data is collected synchronously at a preset sampling frequency and processed with filtering and noise reduction. Using a preset sliding window as the unit, time-domain and frequency-domain related statistical features are extracted from the processed multi-sensor data. The sliding window is measured in seconds, and its duration is set according to the equipment type and actual application scenario.

[0023] In one implementation, this method is applied to the earthwork excavation scenario of urban rail transit construction and is suitable for medium-sized excavators. This scenario is characterized by multiple machines operating simultaneously and frequent vibrations in the surrounding environment of the foundation pit. Excavators often idle while waiting for dump trucks to clear the waste after starting the engine. At the same time, there is a risk that operators may artificially activate ACC to falsely report working hours. Therefore, it is urgent to accurately distinguish between effective excavation operations and ineffective states, resist environmental interference, and prevent fraud.

[0024] Specifically, the multi-sensor system includes a 6-axis IMU sensor deployed at the bottom of the cab to collect equipment vibration acceleration and tilt data; an odometer module installed on the main beam of the chassis to collect displacement and speed data; and a speed sensor fixed in the engine compartment and a pressure sensor in the hydraulic system to collect corresponding operating condition data. The sampling frequency is uniformly set to 1Hz, the sliding time window size is 10 seconds, and the step size is 2 seconds. The sampling frequency, sliding window, and step size of other application scenarios or equipment can be set according to the actual situation.

[0025] First, the synchronously acquired sensor data is low-pass filtered to remove high-frequency environmental noise. Then, the data breakpoints are corrected through a smoothing algorithm. Subsequently, eight statistical features are extracted from the preprocessed data, including root mean square acceleration, velocity variance, tilt angle change rate, spectral energy distribution, displacement increment, mean engine speed, peak hydraulic pressure, and pressure change rate.

[0026] S2. Input the statistical features into the trained decision model and output the state results and confidence scores of each sliding window; where the state results include the equipment state and the pending state, and the equipment state includes the working state and the non-working state.

[0027] Specifically, the decision model uses the statistical characteristics obtained from each sliding window to determine the state of the device. The probability corresponding to the determination result is the aforementioned confidence level, which is a quantification of the reliability of the decision model's determination result. A pending state is a transitional state output when the device state cannot be clearly determined during the inference stage due to the ambiguity and uncertainty of the input features. The confidence level of the pending state is obtained by subtracting the confidence level of the working state from 1, and then subtracting the confidence level of the non-working state.

[0028] In one implementation, a random forest model is used as the decision model. The advantages of the random forest model are its strong resistance to overfitting, good recognition effect of nonlinear feature combinations, accurate capture of statistical feature pattern differences under different equipment conditions, and adaptability to the complexity of data in urban rail transit earthwork excavation scenarios.

[0029] The training process of this decision model is based on the actual operating data of the target equipment. First, multi-sensor data is collected from the medium-sized excavator in this embodiment under various scenarios, including real excavation operations, idling, environmental vibration interference, and artificial shaking. After preprocessing and feature extraction consistent with step S1, sample data containing 8-dimensional statistical features is obtained. Simultaneously, the equipment status corresponding to each sample is manually labeled, constructing training and validation sets. By adjusting hyperparameters such as the number of decision trees and maximum depth in the random forest model, the random forest model is trained and iteratively optimized to complete the model training.

[0030] When the decision model runs inference, it takes the 8-dimensional statistical features of each sliding window as input, and after parallel inference and voting fusion of multiple decision trees, it outputs the state result and confidence level corresponding to the window.

[0031] S3. Verify the state results of each sliding window based on the interference feature library. If interference is detected, the sliding window is determined to be in a non-working state, and its confidence level is adjusted to the preset high confidence threshold. If there is no interference, the state results of the sliding window are not changed.

[0032] Specifically, the interference feature library is a pre-built collection of multi-dimensional feature vectors storing typical interference types in engineering scenarios. Its core function is to accurately match and identify false work signals in the sliding window caused by human manipulation or external environmental factors. The dimension of its feature vectors is completely consistent with the statistical features extracted from the sliding window, ensuring compatibility and accuracy in matching. When interference is detected, the device status of the sliding window is adjusted to a non-working state, and its confidence level is adjusted to a preset high-confidence threshold. Because it is considered a result of interference, the probability of it being a non-working state is extremely high. This application presets a high-confidence threshold to assign to the confidence level of this type of window device status.

[0033] In one implementation, the construction of the interference feature library is guided by the urban rail transit earthwork excavation scenario: First, the high-frequency interference types in the scenario are sorted out, including but not limited to the deceptive behavior of operators shaking the equipment, environmental vibrations generated by nearby construction machinery such as road rollers or dump trucks, false displacements caused by GPS signal noise, and slight vibrations when the equipment is stationary; then, multi-sensor data under various interference scenarios are collected in a targeted manner, and after low-pass filtering, smoothing and feature extraction in the same way as in step S1, the corresponding 8-dimensional feature vectors are obtained, and the interference types are manually labeled to form an interference feature sample set; the interference feature sample set is optimized by clustering algorithm to remove duplicates and retain the typical feature vectors of various interferences, and finally the construction of the interference feature library is completed, and the feature vectors in the interference feature library can be supplemented with new interference types through an adaptive update module.

[0034] During verification, the 8-dimensional statistical features extracted from the sliding window are first normalized to eliminate the dimensional differences between different dimensions. Then, a cosine similarity algorithm is used to calculate the similarity between each feature vector and all typical interference feature vectors in the interference feature library. A matching threshold is set. If the calculated similarity is greater than or equal to the matching threshold, it indicates that the feature of the sliding window highly matches a certain type of interference feature, and interference is determined to exist. The device status of the window is directly corrected to a non-working state, and its confidence level is set to 0.95. If the similarity is less than the matching threshold, it indicates that there is no matching interference feature, and the original state result output by the decision model remains unchanged, accurately blocking false operation signals from the data level. The high confidence threshold is 0.95.

[0035] S4. Based on the sliding window, mark the status result of each time point. According to the misjudgment elimination rule, handle the conflict of status results and pending status at each time point. Accumulate the time points that are continuously marked as working status. When the accumulated duration is greater than the exclusive working time threshold of the engineering machinery equipment, it is included in the effective working time.

[0036] Specifically, multiple sliding windows may cover the same point in time, and the state results obtained by the decision model from these sliding windows may be inconsistent, resulting in multiple possible state outcomes for that point in time. Furthermore, the decision model may continuously output pending states, which need to be clearly defined as either working or non-working states; otherwise, it will affect subsequent work hour statistics. To address these issues, this application proposes a misjudgment elimination rule. This rule ensures that the equipment state determination at each time point is objective, clear, and matches the actual operational patterns of the engineering scenario.

[0037] Different types of construction machinery have different operating characteristics. For example, excavators have long digging cycles and loaders have high loading and unloading frequency. Therefore, it is necessary to preset specific working time thresholds based on equipment type and operating scenario as the benchmark for determining effective working time.

[0038] In one implementation, the time points within the sliding window are marked based on the state results of the sliding window determined by the decision model. For a time point covered by multiple sliding windows, if the multiple state results of that time point are marked in the same way, then its state result is determined; if they are not in the same way, then it needs to be processed according to the misjudgment elimination rule.

[0039] The status results and corresponding confidence levels of all coverage windows at that time point are summarized, and a comprehensive judgment is made in combination with the core characteristics of continuous operation and stable status of engineering machinery and equipment. The status results supported by most sliding windows and with higher confidence are used as the core reference to avoid random errors caused by instantaneous data fluctuations and short-term sensor noise in a single window.

[0040] For any pending state time points that remain after the above comprehensive judgment steps, further analysis of the preceding and following time periods is conducted to clarify the equipment state attribution: If a pending state occurs at a single point in time or at consecutive points in a short period, its attribution is determined by referring to the equipment status of the periods before and after it. If a pending state occurs consecutively and its duration reaches the preset pending time duration standard, it is determined to be a non-working state. For all pending state time points, the attribution of the equipment status is uniformly clarified to ensure that the equipment status at each time point is unique and clear, providing a reliable basis for the cumulative statistics of continuous working time.

[0041] In this embodiment, the sampling frequency is 1Hz, that is, one second is taken as a time point, and the time points continuously marked as working state are accumulated. At the same time, based on the medium-sized excavator in the earthwork excavation scenario, its exclusive working time threshold is preset. When the accumulated continuous working time exceeds the exclusive working time threshold, the time segment is included in the effective working time, ensuring that the statistical effective working time matches the effective working time actually completed by the equipment.

[0042] S5. Periodically evaluate the accuracy of the status results, and adjust the parameters of the decision model and update the interference feature library based on the feedback.

[0043] Specifically, the periodic assessment is based on the actual operating cycle of the construction machinery and equipment. The core of the assessment revolves around the equipment status judgment results and the actual operating situation. By combining feedback information from multiple sources such as construction logs, manual on-site verification records, and equipment operation and maintenance data, cases of judgment deviation are screened out, including actual operation being misjudged as non-working or pending state, non-working state being misjudged as working state, and misjudgment caused by newly emerging unidentified interference signals. The overall judgment accuracy, misjudgment types and proportions are quantitatively statistically analyzed to clarify the direction of parameter optimization for the decision-making model and the need to supplement the interference feature library.

[0044] Based on the feedback results, the parameters of the decision model are optimized: the original data of multiple sensors corresponding to the deviation cases are extracted, and preprocessing and sliding window feature extraction are completed according to the preset process to form a new sample set; the new sample set is integrated into the original training data, and the decision model is retrained iteratively. By adjusting the feature weight allocation inside the decision model and optimizing the state judgment logic, the decision model learns previously unrecognized operation feature patterns or misjudged features, thereby improving its adaptability to the recognition of equipment status under complex working conditions.

[0045] Meanwhile, for new types of interference discovered during the assessment, such as vibration patterns of newly added nearby equipment or artificial falsification methods not included in the database, multi-sensor data under these interference scenarios are collected. After feature extraction, corresponding feature vectors are obtained, and after standardization, they are added to the interference feature library to enrich the coverage of interference features in the library. If it is found that some features in the original interference feature library have decreased adaptability to the current scenario, these features are updated and optimized simultaneously to ensure that interference verification can effectively deal with new and old interference types. This allows the time statistics method to maintain high judgment accuracy and strong anti-interference capability in the long term through adaptive iteration.

[0046] In summary, this application, through ACC signal prior triggering, multi-sensor data fusion and sliding window feature extraction, decision model judgment, interference verification and false judgment elimination protection, combined with dedicated working time thresholds and adaptive update mechanisms, accurately removes invalid states, resists various interferences, adapts to multiple types of equipment and complex working conditions, and significantly improves the accuracy, reliability and universality of working time statistics for construction machinery equipment.

[0047] Based on the aforementioned method for calculating the working hours of engineering machinery and equipment, this application further provides a verification method based on an interference feature library to enhance anti-interference capabilities, including: The system calls a preset interference feature library and compares the statistical features extracted from the sliding window with the interference features in the interference feature library. If the statistical features of the sliding window match the interference features, it is determined that the sliding window has interference; otherwise, it is determined that there is no interference.

[0048] The interference feature library contains pre-stored interference features, including those caused by human manipulation and external environmental interference. The dimensions of the interference features are consistent with those of the statistical features extracted by the sliding window.

[0049] Specifically, the aforementioned interference features are derived from statistical features obtained by collecting, preprocessing, and extracting features from multi-source sensor data of engineering machinery and equipment in actual engineering scenarios. These features are then filtered and validated based on data from falsification behaviors and environmental interference scenarios under real working conditions before being incorporated into the interference feature library.

[0050] The method for comparing the statistical features extracted by the sliding window with the interference features in the interference feature library is as follows: The statistical features are normalized and then similarity is calculated with the interfering features. The similarity calculation uses cosine similarity or Euclidean distance algorithms. If the calculated similarity value is greater than or equal to the preset similarity threshold, or the Euclidean distance value is less than or equal to the preset distance threshold, then the statistical features of the sliding window are determined to match the interference features.

[0051] Specifically, similarity calculation can be performed using either the cosine similarity algorithm or the Euclidean distance algorithm, with different algorithms corresponding to different preset similarity thresholds. The cosine similarity algorithm is used to determine the directional consistency of feature vectors, adapting to interference scenarios dominated by directional features such as vibration and tilt angle; the Euclidean distance algorithm is used to determine the numerical difference of feature vectors, adapting to interference scenarios dominated by numerical features such as load and displacement.

[0052] If the cosine similarity algorithm is used, the statistical features and interference features extracted by the sliding window are normalized to eliminate the influence of differences in the dimensions of different features on the calculation results, and the normalized statistical features are obtained respectively. and interference characteristics The formula for calculating cosine similarity is: , The cosine similarity value ranges from [-1, 1]. The closer the value is to 1, the higher the matching degree between the statistical features and the typical interference features; when the calculated value is... If the similarity is greater than or equal to a preset similarity threshold, the statistical features of the sliding window are determined to match the interference features.

[0053] If the Euclidean distance algorithm is used, the formula for calculating the Euclidean distance is: , The smaller the Euclidean distance value, the lower the difference between the statistical features and the typical interference features. When the calculated Euclidean distance value is less than or equal to the preset distance threshold, it is determined that the statistical features of the sliding window match the interference features.

[0054] Among them, the preset similarity threshold and preset Euclidean distance threshold are calibrated based on different types of engineering machinery and equipment, working conditions and historical interference data.

[0055] In one implementation, the eight statistical features extracted by the sliding window—root mean square acceleration, velocity variance, tilt rate of change, spectral energy distribution, displacement increment, mean engine speed, peak hydraulic pressure, and pressure rate of change—along with typical interference feature vectors of the same dimension pre-stored in the interference feature library, are subjected to min-max normalization processing. This maps each feature dimension to the 0-1 interval, eliminating the interference of dimensional differences between different feature dimensions on similarity calculation, and yielding normalized feature vectors. and .

[0056] Based on the operating scenarios and feature vector distribution characteristics of engineering machinery and equipment, cosine similarity or Euclidean distance is selected to calculate similarity and match. In this embodiment, the Euclidean distance algorithm is used for calculation; the normalized feature vectors are... and Calculate the Euclidean distance separately, and determine whether the sliding window matches the interference feature according to the preset distance threshold in this embodiment. If it matches, it is determined that the sliding window has some kind of interference or human fraud behavior, and the device status of the window is set to non-working state, and the confidence level is set to 0.95. If it does not match, the device status judgment result of the decision model is retained, and the device status of the window is not changed for the time being.

[0057] Because this application uses a sliding window to process data for each time period, and these sliding windows cover the same time point—for example, in this embodiment, the sliding time window size is 10 seconds and the step size is 2 seconds—the 3rd second time point includes both the judgment results of the first time window (1-10s) and the judgment results of the second time window (3-12s). If the two sets of judgment results are different, the device status at that time point cannot be determined. Furthermore, the decision model's judgment results include a pending state, which is neither a working state nor a non-working state. To address these issues, this application further proposes a misjudgment elimination rule to handle conflicting state results and pending states at each time point, including: For each time point, summarize all state results covering that time point and their corresponding confidence levels, and use the confidence level voting method to determine the state result for that time point; If there are consecutive time points that are all marked as pending, and the length of the corresponding time period is greater than or equal to the preset pending time length threshold, then the device status of all time points within that time period will be uniformly corrected to non-working status. If there are single or consecutive time periods shorter than the pending time duration threshold that are marked as pending, the device status is determined based on the preceding and following time periods: When the device status is consistent between the current and subsequent time periods, the device status at multiple time points within a single or continuous time period is set to be consistent with the device status of the current and subsequent time periods. When the device status is different between the current and subsequent time periods, all status results covering the current time point or each time point within a continuous time period and their corresponding confidence levels are summarized, and the confidence level voting method is used to determine the device status of the pending state time point or the pending state time point within a continuous time period.

[0058] The confidence level voting method includes: for each time point, accumulating the confidence levels corresponding to each state outcome to obtain the sum of the confidence levels corresponding to each state outcome, and marking the time point with the state outcome that has the highest sum of confidence levels.

[0059] Specifically, when multiple sliding windows overlap on the time axis, the same point in time may correspond to multiple different equipment status judgment results. At the same time, the decision model may also output a pending state when the features are not obvious or there are conflicts. If not processed, this will directly affect the continuous statistics of effective working hours. Therefore, this application sets a misjudgment elimination rule to uniformly process equipment status conflicts and pending states at the time point level.

[0060] The core logic of the confidence voting method is to eliminate the ambiguity of equipment status caused by overlapping windows by quantifying the reliability of the equipment status determination of each sliding window. The hierarchical processing rule for pending states is based on the threshold division of the continuous pending time length, combined with the correlation of the status of the preceding and following time periods, to achieve accurate merging of pending states and avoid the breakpoints in work hour statistics caused by isolated or short-term pending states.

[0061] In one implementation, after completing the state determination and interference verification of the sliding window, the misjudgment elimination process is performed one by one for all time points in chronological order; for each time point, all sliding windows covering that time point are retrieved, the determination result of the corresponding state result and its confidence level are obtained, and the confidence levels of different state results are accumulated and calculated, and the state result of that time point is determined based on the sum of the confidence levels.

[0062] When multiple consecutive time points are detected as being in a pending state, the duration of this pending state is continuously recorded. If the duration reaches or exceeds a preset pending duration threshold, all time points within that consecutive time period are uniformly marked as non-working. If the duration does not reach the pending duration threshold, the device status before and after the consecutive pending period is further considered: if the device status before and after the consecutive pending period is consistent, the time points within the consecutive pending period are uniformly corrected to the same device status as before and after the consecutive pending period; if the device status before and after the consecutive pending period is inconsistent, the working and non-working statuses and their confidence levels corresponding to each time point within the consecutive pending period are re-summarized, and the final device status of the time points within the consecutive pending period is determined by a confidence level voting method.

[0063] By using the above processing method, each point in time corresponds to a unique and clear equipment status, providing a stable, continuous and reliable status input for the accumulation of working hours in subsequent continuous working states.

[0064] The specific working time threshold for construction machinery and equipment is determined based on the equipment type, operating characteristics, and preset working condition requirements, and is used as a benchmark for determining effective working time.

[0065] Specifically, the dedicated working time threshold for construction machinery equipment is used to limit the minimum duration required for the equipment to be counted as valid working time when its status is continuously determined to be in a working state. By incorporating equipment type, operation characteristics, and preset working condition requirements into the threshold setting criteria, different construction machinery equipment have differentiated valid working time determination standards during the working time statistics process. This avoids miscounting short-term, intermittent, or unstable working states as valid working time, ensuring that the statistical results can truly reflect the actual working time of the equipment under stable operating conditions.

[0066] In one implementation, after determining the equipment status at each time point, the time points continuously marked as working are accumulated using the time point as the smallest statistical unit. When the accumulated duration corresponding to the continuous working status is less than the dedicated working time threshold of the engineering machinery equipment, the time period is not counted as valid working time. When the accumulated duration corresponding to the continuous working status reaches or exceeds the dedicated working time threshold, the continuous time period that meets the threshold condition is counted as valid working time. When the equipment status changes from working status to non-working status or is corrected to non-working status by the misjudgment elimination rule, the current working time accumulation process is terminated, and the working time determination of the next continuous working status is restarted.

[0067] This application also provides a working time statistics system for engineering machinery equipment, such as Figure 2 As shown, it includes: The ACC signal prior module is used to determine the operating status of the equipment based on the ACC switch signal, thereby determining the start and stop logic for subsequent working time statistics. The data acquisition and processing module is used to synchronously acquire and process multi-sensor data according to the sampling frequency, and extract statistical features within the sliding window from the processed multi-sensor data according to a preset sliding window. The decision module is used to input statistical features into the trained decision model and output the state results and confidence scores of each sliding window; the state results include the device state and the pending state, and the device state includes the working state and the non-working state. The interference verification module is used to verify the state results of each sliding window based on the interference feature library. If interference is found, the sliding window is determined to be in a non-working state. If there is no interference, the state result of the sliding window is not changed. The working time statistics module is used to mark the status result of each time point based on the status result of the sliding window, handle the conflict of status results and pending status at each time point according to the misjudgment elimination rules, accumulate the time points continuously marked as working status, and when the accumulated time exceeds the exclusive working time threshold of the construction machinery equipment, it is counted as valid working time. The adaptive update module periodically evaluates the accuracy of the state results, optimizes the parameters of the decision model based on feedback, and updates the interference feature library.

[0068] This application also provides an electronic device, including: One or more processors; Memory, used to store one or more programs; When one or more programs are executed by one or more processors, the one or more processors implement a time statistics method for any of the engineering machinery equipment described above.

[0069] The above embodiments are merely illustrative of the technical concept and features of the present invention, intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly, and should not be construed as limiting the scope of protection of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects. The scope of the present invention is defined by the appended claims rather than the foregoing description, and thus all changes falling within the meaning and scope of the equivalents of the claims are intended to be included within the present invention.

Claims

1. A method for calculating the working hours of engineering machinery and equipment, characterized in that, include: Collect the ACC switch signal of the construction machinery equipment. If it is in the on state, perform the following steps: Data from multiple sensors is collected and processed synchronously according to the sampling frequency, and statistical features within the sliding window are extracted from the processed multi-sensor data according to a preset sliding window. The statistical features are input into the trained decision model, and the state results and confidence scores of each sliding window are output; the state results include the device state and the pending state, and the device state includes the working state and the non-working state. The state results of each sliding window are verified based on the interference feature library. If interference is found in the verification, the sliding window is determined to be in a non-working state, and its confidence is adjusted to a preset high confidence threshold. If there is no interference, the state result of the sliding window will not be changed; The status result of each time point is marked based on the status result of the sliding window. According to the misjudgment elimination rule, the conflict of status result and the pending status of each time point are handled. The time points that are continuously marked as working status are accumulated. When the accumulated time is greater than the exclusive working time threshold of the engineering machinery equipment, it is included in the effective working time. Regularly assess the accuracy of the status results, and adjust the parameters of the decision model and update the interference feature library based on the feedback.

2. The method for calculating working hours of engineering machinery and equipment according to claim 1, characterized in that, The method for verifying the state results of each sliding window based on the interference feature library includes: The system calls a preset interference feature library and compares the statistical features extracted from the sliding window with the interference features in the interference feature library. If the statistical features of the sliding window match the interference features, it is determined that the sliding window has interference; otherwise, it is determined that there is no interference.

3. The method for calculating working hours of engineering machinery equipment according to claim 2, characterized in that, The interference feature library pre-stores interference features caused by human fraud and external environmental interference, and the dimensions of the interference features are consistent with the statistical features extracted by the sliding window.

4. The method for calculating the working hours of engineering machinery equipment according to claim 3, characterized in that, The method for comparing the statistical features extracted by the sliding window with the interference features in the interference feature library is as follows: The statistical features are normalized and then similarity is calculated with the interfering features; the similarity calculation uses cosine similarity or Euclidean distance algorithms. If the calculated similarity value is greater than or equal to the preset similarity threshold, or the Euclidean distance value is less than or equal to the preset distance threshold, then the statistical features of the sliding window are determined to match the interference features.

5. The method for calculating working hours of engineering machinery equipment according to claim 1, characterized in that, The method for handling conflicting state results and pending states at each time point based on the misjudgment elimination rule includes: For each time point, summarize all state results covering that time point and their corresponding confidence levels, and use the confidence level voting method to determine the state result for that time point; If there are consecutive time points that are all marked as pending, and the length of the corresponding time period is greater than or equal to the preset pending time length threshold, then the device status of all time points within that time period will be uniformly corrected to non-working status. If there are single or consecutive time periods shorter than the pending time duration threshold that are marked as pending, the device status is determined based on the preceding and following time periods: When the device status is consistent between the current and subsequent time periods, the device status at multiple time points within a single or continuous time period is set to be consistent with the device status of the current and subsequent time periods. When the device status is different between the current and subsequent time periods, all device statuses covering the current time point or each time point within a continuous time period and their corresponding confidence levels are aggregated, and the confidence level voting method is used to determine the device status of the undetermined state time point or the undetermined state time point within a continuous time period.

6. The method for calculating working hours of engineering machinery equipment according to claim 5, characterized in that, The confidence-based voting method includes: for each time point, accumulating the confidence scores corresponding to each state result to obtain the sum of the confidence scores corresponding to each state result, and marking the time point with the state result that has the highest sum of confidence scores.

7. The method for calculating working hours of engineering machinery and equipment according to claim 1, characterized in that, The specific working time threshold for the construction machinery equipment is determined based on the equipment type, operating characteristics, and preset working condition requirements, and is used to provide a benchmark for determining effective working time.

8. A working time statistics system for engineering machinery and equipment, characterized in that, include: The ACC signal prior module is used to determine the operating status of the equipment based on the ACC switch signal, thereby determining the start and stop logic for subsequent working time statistics. The data acquisition and processing module is used to synchronously acquire and process multi-sensor data according to the sampling frequency, and extract statistical features within the sliding window from the processed multi-sensor data according to a preset sliding window. The decision module is used to input statistical features into the trained decision model and output the state results and confidence scores of each sliding window; the state results include the device state and the pending state, and the device state includes the working state and the non-working state. The interference verification module is used to verify the state results of each sliding window based on the interference feature library. If interference is found, the sliding window is determined to be in a non-working state. If there is no interference, the state result of the sliding window is not changed. The working time statistics module is used to mark the status result of each time point based on the status result of the sliding window, handle the conflict of status results and pending status at each time point according to the misjudgment elimination rules, accumulate the time points continuously marked as working status, and when the accumulated time exceeds the exclusive working time threshold of the construction machinery equipment, it is counted as valid working time. The adaptive update module periodically evaluates the accuracy of the state results, optimizes the parameters of the decision model based on feedback, and updates the interference feature library.

9. An electronic device, characterized in that: The electronic device includes: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a time statistics method for engineering machinery equipment as described in any one of claims 1-7.

Images